Risks and Benefits of Phase 1 Clinical Trials Evaluating New Anticancer Agents: Title and subTitle BreakA Case for More Innovation

The development of new anticancer agents follows a well-established paradigm that progresses sequentially from phase 1 through phase 3 clinical trials. Phase 1 clinical trials are first in human studies of investigational agents and enroll patients with advanced or refractory disease for whom no standard options exist. The starting dose is usually about one tenth of the lethal dose in animals that have been used for preclinical studies of toxicity. Cohorts of 3 to 6 patients are treated at each dose level, and the dose escalation in most studies involves higher escalation steps with decreasing relative increments.¹ Dose escalation within the same patient is not permitted in most studies because of the desire to evaluate patients for cumulative or delayed toxicity at lower doses. The primary objectives of phase 1 studies are to evaluate the safety and tolerability of new agents and to recommend doses and schedules for phase 2 studies of efficacy. Phase 2 studies are usually limited to patients with a specific type of cancer and the usual outcome measure is tumor response (ie, for solid tumors, reduction in tumor volume). New drugs that demonstrate reasonable levels of efficacy are then selected for evaluation in large randomized phase 3 trials in which they are compared (either alone or in combination) with the current standard treatment, using outcomes that reflect patient benefit.

Demonstration of antitumor activity is not a primary outcome measure of phase 1 studies, but both patients and physicians expect benefits from treatment received while participating in them.² Despite explicit information given to patients both verbally and in written consent forms, many patients have high and unrealistic expectations about benefits that may even include cure.² In a systematic review of 211 published phase 1 studies enrolling 6639 patients from 1972 to 1987, Decoster et al³ reported an overall response rate of 4.5% (including 0.3% complete responses). Similar findings were observed in other systematic reviews.^{4 - 5} Evidence of antitumor activity in phase 1 studies does provide directions for further clinical development: all new anti-cancer agents approved in recent years by the US Food and Drug Administration demonstrated antitumor activity in phase 1 studies.

Patients who enroll in phase 1 studies can encounter substantial risks. Decoster et al³ reported a toxic death rate of 0.5% in phase 1 studies; these patients are also subjected to adverse effects from investigational agents, discomfort and inconvenience associated with frequent hematologic and radiologic evaluations, biopsies (sometimes repeated), and multiple visits to the physician or hospital. Therefore, the benefit-to-risk ratio of phase 1 studies has been perceived to be low for patients, and the ethical basis of phase 1 studies has been questioned.⁶

In this issue of JAMA, Roberts et al⁷ report recent trends in the risks and benefits of phase 1 clinical trials of investigational anticancer agents. The authors reviewed published reports of phase 1 studies for single investigational agents initially submitted to annual meetings of the American Society of Clinical Oncology (ASCO) from 1991 through 2002. They identified 213 studies, which enrolled 6474 patients. The overall response rate was 3.8% and the overall toxic death rate was 0.54%. The toxic death rate decreased during the 12-year time period of the study, from 1.1% in the first 4 years to 0.06% in the last 4-year period. Response rates also decreased with time, from 6.2% in the first 4 years to 2.5% in the last 4-year period.

A major concern with interpretation of the results of the review by Roberts et al⁷ is that only 44% of the abstracts describing phase 1 trials were subsequently published. Although the authors concluded that published and unpublished abstracts were comparable, studies with unfavorable outcomes are less likely to be published.⁸ Unfortunately, the authors do not provide information about the rate of publication of articles submitted to ASCO during the 3 time-periods of the study, although one would expect a lower rate of publication of recent studies, given the shorter elapsed time since presentation at the ASCO meeting. A small number of unpublished studies with a relatively high rate of toxic death could impact a major conclusion of this review, since there was only 1 toxic death related to a study agent among 1817 study patients in the most recent 4-year time period. The authors indicate that 47% of trials involved molecular targeted agents and postulated that some of this decrease in the toxic death rate could be due to a lower incidence of lethal toxicity of these agents. While investigation of targeted agents has probably increased with time, the investigators did not provide a breakdown of classes of investigational agents over time to document this trend.

Assessment of benefit to patients from participation in phase 1 trials is difficult. Roberts et al⁷ used tumor response rate as the indicator of potential benefit to patients. As the authors note, molecular targeted agents may inhibit tumor growth, thereby inducing prolonged tumor stabilization and delay in tumor progression instead of partial or complete responses. Therefore, it is not surprising that response rates would decrease with time if more phase 1 studies of targeted agents were performed in recent years. Less tangible are the psychological benefits that patients might receive from participating in phase 1 studies, eg, the feeling “that something is being done,” even among patients who recognize that benefits are rare. The quality of life of patients participating in phase 1 studies may improve compared with that of patients receiving best supportive care.⁶

The findings from systematic overviews of phase 1 studies conducted over the last 30 years are remarkably consistent: the response rate is approximately 4% and the toxic death rate is 0.5%.^{3 - 5 ,7} It is unlikely that the risks associated with the first exposure of humans to a new agent in phase 1 studies can be eliminated entirely; therefore, the question before investigators is how to improve patient benefits without a substantial increase in risk. One of the major limitations of classic phase 1 studies is that a substantial number of patients are exposed to subtherapeutic doses of investigational agents.¹ In the review by Roberts et al,⁷ the median dose given to the first patients recruited to the trials was only 20% of the maximum tolerated dose, which is usually 1 dose level above that recommended for subsequent phase 2 studies; given the narrow therapeutic window of most anticancer drugs, responses are very unlikely to occur in patients treated with such low doses.

Previous studies have shown that the majority of responses in phase 1 trials occur at 75% to 125% of the recommended phase 2 dose.^{4 - 5} To avoid undertreatment of patients recruited for evaluation of lower dose levels of phase 1 trials, the initial phase of dose escalation should be accelerated. This could be achieved by starting at a higher dose than the traditional one tenth of the lethal dose in animals,¹ enrolling fewer patients at lower dose levels,⁹ and allowing dose escalation within the same patient. This type of dose escalation is supported by findings from the review by Roberts et al⁷ in that the median response rate in phase 1 studies with this design was higher than in studies using interpatient dose escalation (5.3% vs 3.2%), while the rate of toxic death was similar.

With an ever-increasing number of molecular targeted agents entering clinical trials, end points based on changes in target expression, pharmacokinetics, and functional imaging should be incorporated into phase 1 studies. For example, evaluation of a drug that is targeted to a membrane-based receptor on cells should ideally include an assay to evaluate changes in intracellular signaling from that receptor, so that doses can be correlated with inactivation of the molecular target as well as with toxicity. The appropriate dose for a molecular targeted agent is not necessarily the highest tolerated; it is the dose that effectively inhibits the molecular target in tumor cells. Unfortunately, recent reviews indicate that the majority of phase 1 studies continue to use traditional study designs, and toxicity remains the primary outcome measure for determining the recommended phase 2 dose.^{10 - 11}

The study by Roberts et al⁷ is a timely reminder that investigators should be innovative in designing phase 1 studies. Only through these efforts will patient benefit be maximized, accrual increased, and effective new anticancer agents brought to clinical practice in a timely fashion.